Emanuel

Exactly, that is the interesting part for me too.
I don’t think the best use of this will necessarily be obvious “POV footage” in the usual action-camera sense. The more interesting uses may be the almost invisible ones: situations where the camera is there, alive, following attention, but not constantly being operated in a way that announces itself to everyone in the room.
Your point about parallax and offset is completely fair. If the camera is too far from the head, then the head direction and the actual camera perspective will never match perfectly. But I am not sure that perfect matching is always the goal. In many documentary or making-of situations, what matters is not to reproduce the exact optical perspective of the eye, but to reduce the burden of manual operation and preserve a more natural relation with the subject.
That is why I find it interesting for BTS, documentary and observational work. Not as a perfect substitute for a shoulder camera, not as a Steadicam either, even though, as a gimbal camera, it belongs to a lineage that goes much further back, including the invention that so impressed Kubrick that Garrett Brown ended up operating his own system on THE SHINING. I see it more as a small autonomous B-camera, or a reactive camera system, able to follow intention without forcing the operator to constantly lift, aim, correct and therefore disturb the situation.
In a film set, for instance, the value may be precisely in allowing the device to become boring after a while. If people stop reacting to it, or react to it less than to a conventional camera operator moving around with a visible setup, then it starts doing something useful.
So yes, I think the strongest uses for it may be almost invisible. Not spectacular POV shots, but the kind of footage that exists because the tool reduced the amount of intervention needed to capture it.

I've been thinking about lenses a lot over the last few years, and just to be cheeky I've put some observations into a framework.
Level 1 is where we start - with zooms
The normal start to using lenses is with zoom lenses, probably the kit lens.
We know the thinking at this stage: its convenient, you stand in one place and zoom, hooray!
Level 2 is what most YT lens videos are about - primes are better than zooms
We all know the arguments.  Primes make you "zoom with your feet", they make you learn about perspective, they're sharper, better in low-light, BOKEH!!!!1, you can learn the FOV and develop an instinct for it, vintage ones are cheap, "real photographers / cinematographers use primes!"
There are approximately 1000 billion videos and tutorials explaining this, but this seems to be where the thinking stops.  I've not seen that much stuff that goes beyond this, but this is really just the start.
Level 3 is where understanding begins - zooms and primes have their uses
Almost none of the discussion up until this point acknowledges that lenses create images, and images have aesthetics, and aesthetics are what is actually being discussed.
Moving to talk about motion pictures now, and cinema especially, there is a bunch of nuance that Level 2 doesn't really discuss.
People have decided that FF sensors are the most 'cinematic' and typically are used with FF vintage lenses.  This means that the FOVs are 24mm / 28 / 35 / 50 / 85 / 100 etc, with maybe a 40mm in there if you're getting fancy.
These weren't the FOVs of cinema though, because cinema was S35.  So the FOVs of cinema using the 50/40/35/27/18 were really like 75mm/60mm/52.5mm/40.5mm/27mm.
It gets stranger when you add anamorphic into the mix.  If I go to B&H anamorphic cinema lens category and sort by best sellers, we get:
 - DZOFilm Arcana Anamorphic Prime 3-Lens Kit, which are FF and 32/45/75mm and 1.5x, so on FF they are: 21mm 30mm and 50mm
 - BLAZAR LENS Talon 50mm T2.1 1.5x, which is FF and equivalent to a 33mm
 - Sirui Saturn 35mm T2.9 1.6x which is FF and equivalent to 22mm
If you're using the standard FF lenses on a FF camera, you are using the FOVs that stills photographers used, rather than those that cinematographers used.
Shooting on S35 sensor size (or crop mode) with FF lenses can create some of these in-between FOVs too.
People at Level 2 thinking probably won't be swayed by the above.  I would imagine the thinking is take a step forward or back, what's the difference?
Level 4 is where understanding begins to mature - enter the feedback loop
The feedback loop is where you realise that the focal length changes how you shoot.
A ridiculous example to illustrate it.  You decide to shoot on only a 28mm on a FF camera, but when you frame up a close-up shot, the distortion makes the talent look awful, so you take a step back and now the footage feels more distant because we're not seeing the talents face so much because there are no close-ups.  
We all know about perspective from level 2 thinking, but the level 3 thinking was that taking a step forward or back was no big deal, so which is it?  This stuff is subtle, but (like all feedback loops) it pushes us to act differently and this can create a cascade of changes over time.
Level 4 thinking realises that this dynamic is powerful and pervasive.
I shoot in public, so I don't control the environment.  I discovered that if I shoot with a 35mm FOV then I can get environmental portraits of my friends and family from close enough that people won't walk in-between me and them, but moving beyond a 45mm I'd either get shots of them that were tight and didn't really show their environment that well, or I'd step back and be struggling with people walking in-between me and the subject, which is a completely different situation.  How would I respond to this?  I might shoot from eye-level instead of chest height.
Now I've changed the shot angle because of a FOV change.  
If I shot from eye-level for a while I might notice that I get more attention and now I find that the people interacting with my subject are more aware they're being filmed and keep looking at the camera.
Now my subjects are acting differently because of a FOV change.
If I asked someone the difference between shooting with a 35mm and 45mm would they think it would change the shot angle and subject behaviour?  Not with the Level 2 thinking of "primes are sharper! duh!!", or the Level 3 thinking of "just take a step back! duh!!".
What about controlled sets?  
Sure, on a controlled set there aren't random people walking in-front of the camera, but now we're talking about actors and all the dynamics that goes on there.  
Can great actors deliver amazing performances while the matte-box is only inches from their face?  Sure.  Do YOU have actors that are that good?  I don't think so.
Can great production designers change a set to accommodate a camera being further away, while keeping the frame looking the same?  We know that as we move the camera back the subject gets smaller in frame, and that as we do that the background gets smaller but not nearly as fast as the subject does.  This is great if you are only filming the subject and don't really pay attention to the composition of the entire frame.  But you're a talented cinematographer, so you want to move back a bit and keep the same composition, which means that production design needs to 'cheat the camera' and basically rearrange every item in frame that isn't in the very background.  I remember shooting a student film in a cafe and every setup required moving the vase of flowers on the table the subject was sitting at.  That vase probably used two-thirds of the area of the table!
I watched a video recently where a street photographer tested a 40mm prime for the first time.  They didn't know what to make of it, having only a week to shoot with it before they had to release their video review.  What struck me wasn't that they didn't know what shooting with a 40mm was like, it was that they didn't seem to understand that there's a period of learning that goes on, they didn't understand that the feedback loop exists.  
I realised they had 'learned' each focal length by memorising its attributes (which Level 2 photographers will crap on at great length about), rather than having learned them for himself by following a process where you explore the feedback loop and see how it makes you feel and how it makes you act and how the world responds to that, and how you respond in turn, and how the loop feels and matures over time, and how to make the loop go faster etc.
I recently spent some time in a small town in rural Japan and shot the same location with FOVs equivalent to 71mm, 82mm, and 100mm.  I went out for a walk each night with one of those lenses, going out for perhaps an hour or two.  Shots that were possible with one were not with the next, shots that were great with one were lifeless with another.  As I walked down the same road from my accommodation seeing the same shots night-after-night and making different framing decisions with each lens (and deciding to take the shot or not to bother as it didn't work) I noticed that I made different decisions to walk one way or another as certain subjects required different FOVs and distances to make them.
I've also spend a lot of time, over several trips, shooting night scenes with 68mm and 71mm FOVs.  In some locations I can make some shots and not others, while in other locations I can take different shots.  If I'm shooting across a road then the width of that road (combined with my FOV) determines the type of shots I can take.  After taking a number of those types of shots I start to adapt to how I'm shooting these locations.  The more I shoot the more everything feels different.  Level 3 thinking says "just take a step back, what's the difference?" and when shooting in those situations the difference between a 35mm and a 50mm feels like it's a span where there are several complete aesthetics in-between the 35mm end and the 50mm end.  Thinking about shooting a 50mm FOV vs an 85mm FOV feels like travelling to a different country where things look similar but feel very different in practice.
I know I'm barely scratching the surface of Level 4, and perhaps there are levels beyond this that I'm unaware of, but it's just amazing to me that almost no-one seems to talk about anything beyond Level 2.  It's probably controversial to say, but I deliberately avoid almost all stills-only people because the thinking seems so rudimentary in comparison to people who shoot moving images.  You can feel the limited thinking and the "well, actually!!!!" responses where they miss the entire point entirely because one lens is sharper or something ridiculous.
Anyway, hopefully this helps.  I've not really heard anyone talk about this stuff, which seems a shame as the rabbit hole is very deep and to only talk about ankle-deep water seems silly.

I'm once again reminded of Noam Kroll, who has gone a long way into this rabbit hole.  
My recollection of his method was a balanced approach, where you make a plan and then improvise and adapt within a limited range.  My impression was that he would storyboard things as a way of mentally rehearsing the shoot, and would end up with a clear idea of the logistics of the shoot, the equipment required, etc.  Location scouting and anticipating the light etc as you normally would.
I believe he also gained a clear idea of which shots were required, and which had some flexibility.  Then when he was shooting he could make sure that he got enough for a functional edit, but was also clear enough in his thinking that he could adapt the plan to compensate for any challenges that arose and also to take advantage of any serendipity or inspiration that occurred.
I suspect that this is a very deep skill, to plan and then improvise a shoot with an understanding of how the choices being made will go together in the edit.  I know enough about editing to know that it's a jigsaw puzzle where you can have two small sequences that work well but don't cut together directly, so unless you can find a way to get from one sequence to the other then you have to change one of them so they're compatible.  To do this for a whole scene, or whole film, in your head while you're still shooting it is beyond what I could even imagine, but I'm sure that the talented cinematographers are easily up to the challenge.
Noam actually went further, describing a process where he worked with two actors and where he 'designed' what would be shot ahead of time, with the major plot points and story beats, but didn't fully script it.  On each day of shooting the three would have breakfast and discuss the motives of the characters and how the scenes should go.  Then they'd shoot while improvise the scenes, filming as they went and exploring ideas.  It was freedom within a planned structure.  
I believe he was shooting one or two days a week, and so after shooting he'd review the footage and do rough edits, seeing what worked and what didn't.  Then he'd 'design' the next shoot day accordingly, sometimes keeping on with his overall plan for the story but other times seeing something in the footage that made him adapt the narrative.
I suspect that the skill is in knowing how much you can stray from the plan and knowing in which ways to adapt to make the end result better than if you just shot it as planned without any adapting to the situation.
Certainly if you make a plan and then prioritise which shots are the most to least important then you'll have a good chance of coming back with a functional edit.  My impression of great travel content is that most shots are good-but-not-great, and the art is in the edit and how they're combined.
EOS-M and primes and a Fujinon-TV 14-70 f/2 will do a lot of the heavy lifting in making it look like cinema instead of video.  In my mind you'll need to pay attention to how to keep the camera stable and then look at your references and study their coverage so you can design yours.  
By shooting on less than pristine equipment, you'll have to get things right in-camera as you won't be able to mess with it in post as much.  Specifically, being able to zoom in a little in post can be useful if a random passer-by is staring.  If you were shooting this with a modern mirrorless and sharp/neutral lens then I'd suggest using the highest resolution possible just so you have that flexibility.  
Your concern for getting stared at is legitimate, but the focus is to not get people staring while they're in the frame.  As such I'd suggest getting more coverage using tighter framing and shallower DOFs, and for shots that are wider, simply getting more footage so you can edit around people staring.  AI can potentially help if there are random people staring in shots you really want to use, but if you can edit around these moments (or prevent them from being in shot in the first place) then all the better.
It's also worth considering that there are a number of things you can do that will lessen the changes of people noticing you and the camera, or lessen the people who are currently in frame noticing it.  
Another strategy is to investigate how much b-roll can be used in the edit without it taking away from the story.  You may be able to get away with putting b-roll on top of a good audio edit, essentially having an L-cut followed by a J-cut where the audio goes from one character to the another but the visuals go via a b-roll shot from the location.  I'm sure there's a deep art to doing this, but it's worth grabbing as much b-roll as you can while on location (especially if shooting has to wait for any reason but you're able to shoot).  You can even return at a later time to get more footage, or better yet, take your kit and go shoot the location ahead of time so you can do a dry-run with the actual rig and also get a sense of what the location is like to shoot by actually shooting in it.

So far, we have:
The French Connection The Day Of The Jackal Escape From Tomorrowland Chubby Rain Woody Allen (Annie Hall, Hannah and Her Sisters, Deconstructing Harry) Lost in Translation I think this is a great list! I know that Sean Baker’s Tangerine was shot this way, but I’d rather not sit through another of his films.
 
I've written a drama short. Two characters, 11 pages, lots of walking around town. I live in Portland and I’ve shot all over town before without ever once being hassled (except by some in a crisis of mental health). But I’m really more interested in seeing what kind of aesthetic choices might even be available to me when balancing that with coverage needs, moving quickly, and not attracting too much attention.
I'm always so envious of New Yorkers for this very reason. You can fill the frame with sidewalk commuters and often no one will even bother to give you a second glance. I can’t think of any other US cities where this is true. 
Lovely screen grabs, Kye. If any of these were publicity stills for a film, I’d want to check it out. Especially the 68mm shots. Great mood and sense of place.

I think that this quote sums up some of the decision-making that I am facing. I could totally go in there and “wing it”, just figuring out my shot as I went. But I want to rely on natural light, which means timing certain shots for certain times of day, and then that raises the question of how much planning I want to put into each scene, and ultimately how the visuals will compliment the story. I’d rather not lean too far into the documentary aesthetic.
After shooting my last five shorts with my C70s, I’m actually thinking about shooting this film on a hacked EOS-M with c-mount lenses. I’ve been collecting them for a while and have three sets of primes, plus some great zooms. I’d personally put the aesthetic available with this (windowed) sensor and these lenses up against anything, I like it that much. I think the reason that no one really wants to do this is because it’s one of the most difficult digital cameras to work with, AND on top of that, the most difficult lenses to work with. But I’ve been tackling the challenges presented by these one by one, and I think I have them pretty well sorted. We shall see!

Current plan is to hide a Deity PR-2 recorder with a lav on each actor and attach a timecode box to the audio input of the camera. Stabilize with a steadybag. Transmit the image wirelessly to a producer and make up artist in a car, along with refreshments for the actors. If I end up working with long zooms, I might give direction over a phone and provide a discrete earbud to each actor so that I could also monitor.

Further to the above, and further to what Mercer wrote..
The smaller the camera package the more amateur and less pro you look, which impacts how the authorities treat you Often locations care if you have a tripod or not, especially in crowded situations where a tripod takes up a lot of space and is a tripping hazard.  Alternatives to a tripod are obviously hand-held and also shoulder-rig, but the often overlooked options are a monopod, and Mercers trick of having a monopod where the foot is resting in a pocket of a belt, so the camera effectively gains the stability of the operators waist Depending on the focal length and type of shot (medium, close-up, etc) the primary consideration in crowded places is if people will walk in between the camera and subject.  My travel shooting with my family was done mostly on a 35mm F1.9 equivalent and this enabled medium-close-ups and closer in very crowded places without anyone getting in-between and wider than that with people or obstacles in-between.  If you want to get more distance than that and not get wider then you'd need to go to a 50 or tighter depending on the distances involved.  I personally find this hugely situationally dependent as it depends on how crowded things are, how noticeable the operator is, how willing to walk in front of a camera people are (or how much/little they care about you) etc. Combined with the distance / density / shot-size / focal length interactions are the DOF considerations, specifically how much do you want to separate your subject and at what distances.  Normally this also blends into low-light requirements but I think these days if you use a dual-ISO camera then that consideration drops away and you can get by with F2.8 or even F4 at night in well-lit areas.   The main reason that aperture isn't an obvious choice (just go F1.4!) is that if you can choose a slower lens then you can consider a zoom, which changes the shooting equation drastically.  Depending on how you're planning and scheduling the shoot, the ability to move fast without changing lenses might be considerable.  The French New Wave approach of getting minimal coverage and preferring longer takes is something to consider.  There's a huge difference in logistics between storyboarding the whole thing within an inch of its life (and having many setups and doing hair/makeup/wardrobe touchups between takes etc) and running the whole scene a couple of times with a wider master then going a bit tighter and grabbing the more interesting shots as colour for the edit.  Noam Kroll has shot short films on film and only had ratios of 2:1 or similar, and for certain sections only shot one take because he wanted to spend more film on making the important parts more interesting. For aesthetics it's also worth considering what you'll do in prod vs post.  The traditional prod approach is to use filtration and select a lens / aperture combination that gives the rendering you want, and then you'd shape the light and control your lighting amount and ratios etc to suit your ISO/aperture/filtration.  This makes prod very cumbersome and if you don't control the location perhaps impossible.  The alternative approach is that you choose much more neutral equipment and push a bit of a look in post.  There are obviously limits to this, but for example by picking a lens that's sharper across its range you can vary the aperture to control DOF and exposure in prod and then degrade it in post (soften it globally and in the corners, add distortion, add diffusion, add vignetting, etc) and you'll have a consistent look despite using the lens at different apertures, etc. Think about DR.  The more DR the camera has the less of the scene you will clip and the more flexibility you'll have to adjust exposure and ratios etc in post without making the clipped areas visible.  The less DR you have the more carefully you'll have to expose, and the less flexibility you'll have with moving shots that go between dark/light areas.  The more DR you have the less you need to vary the aperture on the lens to compensate, or the less you'll need any lighting etc to compensate. Think about the contrast of the final film.  The more contrast you apply, the more leeway you will have with the cameras DR, so the previous point gets easier.  Film was great in this sense as the negative was so wide and flexible and gave a lot of leeway in post. Monitor as well as you can.  Use a large monitor and a viewing LUT.  The more you can visualise the end result while shooting the better.  I find that shooting in uncontrolled situations means there are always things in the frame that I'm reacting to.  This is in alignment with the situation and performance too - shooting in crowded public places will have the cast reacting to their surroundings, so you should be reacting to their performance and to your surroundings too, so the more clearly you can see the shot the more coherently you can react to it. Embrace the chaos.  Separate the ideas of controlled coverage and creative experimentation as much as you can.  The idea of getting a master in the can and then experimenting is great because you can ensure you've got an edit that can work and then you can grab risky but potentially great shots after that.  Much better to have the final edit cut between neutral shots and really great shots that embrace spontaneity and add to the film than struggling in the edit by having to cut between shots that are neither safe nor creative nor sensitive to the surroundings. Some example 35mm F1.9 shots I've taken (please ignore the grading - these were from a long time ago!!!):




More recent shots with 68mm F1.5 equivalent:




and more recent with 70mm F2.0 equivalent:







If you really wanted a minimal set of focal lengths, I'd suggest a 28mm for wides and ultra-packed situations, a 'normal' lens in the 35-50mm range, and a longer one in the 70-100mm range for shots where you are at some distance and don't want a wide.
Your aesthetic should really begin with the emotional arc of the characters in the film, filtered into scenes, then the equipment chosen to express the intended aesthetic while shooting in the specific circumstances of the location and logistical assets and challenges. 

Lost in Translation was famously shot in Tokyo without official permission.  They shot in public with a very minimal crew and moved fast to try and keep ahead of the authorities.  They chose this approach primarily because it was almost impossible to get permission to film there at the time. 
I saw a great doco about the making of it but it's been removed from YT now so can't share it.
I don't know what sort of info you want to know to prep for your film, but there are snippets of BTS online if you dig.  This video shows a bit of BTS from on location (linked to timestamp):
From what I can remember / piece together:
shotonwhat says it was shot on Kodak 320T and 500T using Aaton 35-III Camera and a Moviecam Compact Camera with Angenieux and Zeiss Super Speed Lenses they moved fast to stay ahead of the authorities the cast and crew when out shooting in public was only a few people (camera, sound, director, and talent and I think that's it?) and were all non-Japanese people, and if anyone official came to tell them off the they would just be apologetic but use the language gap to effectively prevent any communication.  They had a Japanese fixer who stayed a distance apart from the group (so they wouldn't be noticed by the authorities) but that was helping with logistics etc and could step in if the situation required it they had challenges with locations (link to timestamp) another snippet of them on location - tripod but not clear if they're using any lights What information are you looking for specifically?

What are you trying to do? As long as you don't have a big rig with lights and a shotgun mic, then you should be good. I've walked around Philly and NYC and see plenty of people shooting video everywhere with gimbals and even tripods sometimes... although I wouldn't recommend that. The key is to look like a tourist and find locations where there aren't many people passing by, or it's so busy, people don't even look twice at you because you look like a tourist. And be prepared to move to a different location if you have to. Philip Bloom, hardly a narrative filmmaker, but his use of super long lenses gives him the freedom to be a little far from the talent. Separation between camera and talent can be a good thing if you're trying to steal shots. 

The Guerilla Film Makers series of guides is getting on a bit but offers some solid insights into zero budget that are still very relevant.
https://guerillafilm.com/books/

Can you think of any narrative films that were shot with natural lighting minimal crews in busy areas/cities? I’m doing some research for an upcoming zero-budget project, trying to figure out a solid approach to the aesthetics. I really want to get out of a locked set for this one.

Interesting points and I definitely see the challenge and how this potentially provides new potential solutions to some aspects.  
@eatstoomuchjam mentions having something mounted on the brim of a hat, and this does in some way create an alternative option.  
Perhaps the added feature of the head tracker is that you can track the direction of the head without having the camera on the head, which makes me question where would you have it?  The further from the head you put it the more parallax error, and unless it could have some sort of offset (which is probably possible) then the framing might be off.  
Maybe it's better as a more walk-around camera rather than if you're stationary and what you're filming is mostly stationary.  I do see "those shots" in docos from time to time where the footage you're seeing wasn't shot well but is pivotal to the story and therefore included in the edit.  For example when crossing a border into a war zone the audio is of the border guards talking with the producer and the visual is the camera seeing the back of a car seat, or everyone is running while chaos ensues, etc.
I can definitely see that it's an alternative to a chest-mounted action camera where the ultra-wide is capturing far too wide a FOV, and where the head-mount would capture the action with a much tighter FOV.
I can also see someone doing BTS rigging one up with an NPF battery and mini desktop tripod and body clamp and just wandering around with it constantly turned on but not always recording, and the fact that it's always on and always moving would mean it would become invisible after a while, making the footage it captures much less influenced by its presence.
Interesting stuff.  I guess it's probably a thing where the right uses for it are almost invisible.  Like how shoulder-mounted ENG / doco cameras can be used to create footage that doesn't call attention to itself, but in certain drastic situations such cameras do really call attention to themselves, this would be able to capture footage that doesn't call attention to itself in situations where no other camera setup could.
Great discussion!

I think the POV Head Tracker should not be seen merely as a vlogging gimmick or as another accessory for people who want to film themselves walking down the street.
There are many different ways of filming, framing and capturing reality. The most “professional” approach is often understood as the most controlled one: you plan the shoot, you discuss the framing, you block the scene, you decide where the camera goes, you decide what the subject is supposed to give you, and then you execute. That is obviously valid, and it is the basis of a lot of good cinema.
But it is not the only way to make images.
There is also another tradition: a more intuitive, observational, physical and spontaneous way of filming, where the camera is less a machine imposing a pre-decided frame on the world and more an extension of the filmmaker’s presence inside that world.
That is where I think something like the Insta360 POV Head Tracker becomes interesting.
The question is not only “what can it do technically?” The question is: what kind of relationship with reality does it allow?
When you are operating a camera in the conventional way, you are always doing several things at once. You are looking, framing, correcting, adjusting, deciding, reacting, and at the same time you are also visibly present as “the person filming”. That presence changes the situation. It changes the people in front of you. It changes the rhythm of what happens. It can intimidate, formalise, freeze or theatricalise reality.
In documentary, this is especially important. The more you plan, the more you risk fixing the subject before you have really encountered it. You may think you are observing reality, but you are already working on a construction of reality. You are no longer only receiving what is in front of you. You are fabricating a gaze, and the subject starts to exist inside that fabrication.
Of course, complete objectivity does not exist. Every image is already a point of view. But there is still a big difference between a camera that constantly announces itself as an intervention and a camera system that allows the filmmaker to remain more physically and psychologically inside the situation.
This is why the POV Head Tracker interests me. It may allow the filmmaker to film without constantly “operating” in the traditional sense. The camera can follow the natural direction of the filmmaker’s attention. The image can become closer to a lived point of view rather than a pre-composed shot. Not perfect objectivity, obviously, but perhaps a more immediate form of subjectivity.
That distinction matters.
A head-tracked gimbal camera could be useful not because it replaces deliberate cinematography, but because it opens another mode of acquisition: a more instinctive, embodied, less intimidating mode. It lets you be present with the subject while still filming. It can reduce the gap between seeing and recording.
In that sense, I see a possible historical parallel with what happened when smaller 16mm cameras became available. Those cameras did not simply make cinema smaller. They changed the grammar of cinema. They allowed filmmakers to move differently, to follow people differently, to enter rooms differently, to film streets, faces, accidents, gestures, private moments and unstable situations in ways that would have been much harder with heavier, more industrial tools.
You can connect that to cinéma vérité, direct cinema, the Nouvelle Vague, the New American Cinema, Jonas Mekas in New York, underground and independent filmmaking, and later the influence of that freer, more mobile language on figures like Cassavetes, Scorsese, and the whole post-studio generation. Even mainstream cinema eventually absorbed some of that looseness, that handheld energy, that search for immediacy.
Jonas Mekas is not just an abstract reference for me. I had the privilege of knowing him personally in the mid-1990s, at the Figueira da Foz International Film Festival, which he used to attend. In that same context, I was also fortunate enough to receive an award as best daily press film critic. More importantly, a project I am still developing today was born precisely from that contact with Mekas. So when I refer to him here, I am not only invoking a name from film history. I am also referring to a very concrete personal encounter with a way of understanding cinema as diary, presence, immediacy, memory and life.
The technology did not create those artistic revolutions by itself. But it made certain gestures possible.
And when a tool makes a new gesture possible, it can also make a new kind of cinema possible.
That is how I would look at the POV Head Tracker. Not as “AI tracking for creators”, but as a small step toward a different relation between body, gaze and camera.
From the end of last month and carrying into this June, I worked on the making-of for A NOITE, Leonel Vieira’s film adaptation of José Saramago’s homonymous play. During the shoot, we used the Osmo Pocket 3 alongside other cameras, including a Sony A7S III, an FX30, a Panasonic GX80/GX85 and other small-format tools, Insta360 included. The film itself was being shot on two ARRI cameras, so this kind of low-profile equipment was obviously not what people on a cinema set are most used to seeing. Even Leonel Vieira, the director, looked at the Osmo Pocket 3 and jokingly said it looked like a toy camera. But that was precisely part of the point. I took the initiative to use it without hesitation, accepting the risks of bringing that kind of device into a professional film set, and combining it with different optical tools, including black mist filters to create atmosphere, Sirui anamorphic lenses and other accessories. In that context, I became very aware of how valuable it would be to have a device that lets me film without constantly managing the camera as an object. Not to mention that 10-bit Log recording is now available on the Luna Ultra as well.
In a making-of situation, the best moments often happen before people know they are “performing” for the camera. They happen between takes, in hesitations, glances, silences, rehearsals, small gestures, private exchanges, and moments when the machinery of cinema briefly becomes human again.
But the moment you raise the camera, adjust the frame, move closer, correct the angle, ask for space or visibly operate, you can lose the very thing you were trying to capture.
The reality in front of the lens changes because of you.
And that is fundamental in a making-of context: to be as minimally intrusive as possible, so as not to disturb the set of the main film being shot. The reactions of the professionals involved are, in many ways, the real subject of a making-of, and those reactions should not be manipulated by the visible presence of the image-capturing device itself. In our case, we were working with a very small crew: two to three people at most. In fact, it was necessary to convince Leonel Vieira to accept a maximum of three people, because ideally he preferred two, and sometimes only one person could be present. In those situations, when only one person was shooting and I still needed two possible angles, a camera A and a camera B, the only viable option was to have a B camera as autonomous and unobtrusive as possible, which is exactly how the Osmo Pocket 3 was used. With a device such as the Luna Ultra and its POV Head Tracker, that kind of work would become much easier, not only during the shoot itself but also later, when reaching the post-production suite and needing more options in the edit.
So a device that lets the camera follow your attention, while your hands and your body remain less occupied by the act of filming, could be extremely useful. It could allow the operator to be less intrusive, less theatrical, less visibly extractive. It could make the camera feel less like a weapon pointed at reality and more like a witness moving through it.
That does not mean this is for every situation. It is not a substitute for composed cinematography, lighting, blocking, lenses, or intentional mise-en-scène. But it could be very valuable for documentary, making-of work, rehearsal footage, street filming, travel, observational cinema, and any situation where spontaneity matters more than perfect formal control.
The professional instinct is often to control everything.
But sometimes cinema gains power when we control less.
Sometimes the most authentic image is not the one we planned best, but the one we were able to receive before reality became aware of our plan.
That, to me, is where the POV Head Tracker could become genuinely interesting.

Well stated @eatstoomuchjam @kye :- )

Well stated @eatstoomuchjam @kye :- )

Fascinating to say the least : ) His text and voice included ;- )

The below video titled "Retro camcorders are EASY to use in 2026 (full guide)" was posted two years ago and has 241K views.
It opens with these shots:



and his first line is "retro camcorders are freaking awesome".
Maybe you're right that getting worse image quality isn't the primary motivator, but if it wasn't then I don't really know how to explain why anyone would be attracted to these camcorders when new/modern ones have all the same advantages of being a single-use device, have great ergonomics, novelty value, etc, but are also far easier to find, far easier to use (and don't require legacy computer interfaces!), are far more reliable, and have much better image quality.
Maybe the draw is that they're old, and therefore that's the novelty, but I've heard people gush over how they love the JPEG look from cameras that have 2MP cameras and the JPGs are hugely compressed and full of artefacts etc.  
Here's another one comparing Hi8 vs MiniDV on the basis of the image alone..  24K views!
It might also be a nostalgia thing, where the image quality is desirable because it's poor, but in exactly the right way.

I’d have urged caution in buying the Luna before DJI released the Pro version of the Pocket 4 anyway but now the P4P has been officially released in China then I would definitely hold out for some comparisons.
One of the major USPs of the Luna is the excellent removeable remote screen but it seems DJI have an interesting solution with a small external remote for the P4P.
Its a bit of a swings and roundabout in that it is more convenient with the Luna but it’s equally true that if you drop the remote then it’s game over until you can locate a new one.
Insta threw everyone a curveball (shills included) by releasing the Luna far earlier than expected and it’s worth considering that this was in part motivated by them discovering their major USP was about to be countered with the P4P solution.
Oh and the 17 stops of dynamic range of the P4P is interesting too.
 

I find it difficult to believe that worse quality as such is the motivating factor, but a less polished result, a less artificially processed and perfected result may be desirable when one wishes to appear authentic and I do believe a lot of people are tired of the ultra-processed images from mobile phones and editing apps. They may also be tired of commercial images for partly similar reasons: DYI images may look more real and home-made, and somehow more true to the person in the photos, even if not captured by extended arm holding the camera but someone who truly knows the person in the picture. In some cases,  commercially produced portraits which often reflect the photographer's tastes and some product or image style that the photographer has found successful and applies to all their clients. Doing it yourself for an authentic feel doesn't mean the quality of the photo has to be poor. Of course, it's possible that some people specifically want a "different" look such as 8 mm film etc. but I don't think this is common or at the core of the issue, the excessive processing and manufactured "perfection" is much more likely to be what triggers a change in fashion (or perhaps I am just wishing that). You may be correct that the smartphone generation sometimes just wants to use something other than their smartphone, that's perfectly understandable and would be a healthy development. 
In my mind "high quality" and "smartphone" are difficult to put in the same sentence with a straight face. High quality for some things, yes, but for a lot of things, not good quality at all.

Not only the Leica :- )

Great posts, sticky this! Your contribution here @kye is priceless and a fine example for everyone, myself included : ) It's always a pleasure to read your thoughts! :- ) Keep going… I’m linking to it elsewhere, BTW ; -) Food for my trainee students. : D

Not only the Leica :- )

I'd use a little known plugin called Tilt Shift Blur (TSB), which comes with Resolve but is very special in a critical aspect.
Normally if you have a node and give it a key then the node calculates things as normal and then uses the key as a transparency effect, so if you used a large Gaussian Blur and gave it a key then you'd get a huge blur mixed with the sharp image at the level of transparency the key dictated.  However, with the TSB, the key defines the size of the blur, so you can vary the size of the blur that way.
For this purpose I'd give it a luma key of the image and adjust the contrast and amount to control the relative amount of blur between the lighter and darker parts of the image.
The TSB is what I use to soften the edges of the frame in my lens emulation nodes, which allows there to be no blur in the centre and it gradually transition to having a larger and larger blurring towards the edges.
The fact that the key input acts as a transparency control really doesn't make much sense when applying most OFX plugins and I'm surprised they haven't made more of them smart like the TSB one where it uses the key as an input to control one or more of the OFX parameters.

There are so many variables when it comes to how you're viewing the film images.
Negative film has wide dynamic range and soft highlight rolloff.  Positive film has much more limited dynamic range and pretty hard highlight rolloff.  Faster film tends to be grainier.
Filmmakers with a big budget would be choosing their film stock for aesthetic/style reasons.  Imagine shooting Taxi Driver on the same technicolor low-grain film stocks that were used for The Sound of Music.  Bright saturated colors would have been terrible for Taxi Driver.  Scorcese chose less gritty films than some others might, but Travis Bickle lives in a relatively desaturated/dark world and that's for the best.
Filmmakers with low budgets were likely to choose the cheapest film stock they could and some even used the leftovers that weren't exposed from the productions of others.    Or in the case of John Waters, whatever film he could steal.
Next, as you said, for these classic films, you aren't necessarily looking at scans from the master negatives.  You might be looking at scans of the release prints.  They didn't always save the masters.  It could even be a second or third-generation print.
Then to add to that, the way the film gets transferred matters.  Did they scan the original negatives or a print?  How was it scanned?  Was the film being scanned perfectly flat?  What compression was used on the scanned image?  Was it scanned or telecine?  If telecine, which projector lens was used during the telecine process?
As far as the lenses, razor sharp lenses have been available for a long time, including in the 50's, and including wide angles.  Lots of vintage wide angles are a little softer in the corners, but they can be very crisp in the center...  but fashion applied in many eras of film, just as it applies now.  For some of the softer images, especially close-ups, they might have been using a net filter, made more complicated by the net filter potentially being mounted behind the lens instead of in front.
https://www.provideocoalition.com/the-secret-life-of-behind-the-lens-nets/
I'm sure I'm forgetting more things too.  Like almost anything going through an analog to digital process, there are about a bazillion variables to consider along the way.

Now I have mostly progressed my GX85 Super-16mm "conversion" I am turning back to a more generalised look at film and The Aesthetic as it pertains to cinema.
My next step is to convert my S16 emulation into a 35mm emulation, which shouldn't be that hard as it's the same stuff but just using more of it, so turning down the grain and backing off the softening and size of halation and bloom etc.
In order to get my bearings I've collected a bunch of frame grabs from cinema over the decades, and some fascinating things have emerged.
Observation One: Film doesn't look like it's gotten cleaner and sharper
This is what people say, but when scrolling through the references, you can find things like this from 1952's Bend of the River:

and then things like this from 1994s Speed:

Obviously you need to be smart about things, so both these shots are likely to be locked off, focus wouldn't be missed, lit naturally and exposed properly, etc.  There are lots of shots from Speed that are soft, but it's an action movie so lots of them are probably motion blurred or the action was impacting how well that frame was focused etc.
Luckily, License to Kill from 1989 will get us back to safe ground with some nice sharp images:


But how are these images possible all the way back then?
One thing that comes to mind is that when consulting Kodaks excellent Chronology of Film page, you see that the negative stocks started off as very low sensitivity and increased over time, so it's like comparing the high-ISO of todays cameras to the native ISOs of past cameras.
Observation Two: Images look like they've gotten less worse
The further back you go, the more you find shots that should be sharp but just aren't.  The example above from 1952 was a real outlier, as most of the images from that time looked more like these from 1955s A Bad Day at Black Rock:



I suspect the quality of the lenses.
All the above looked like wider shots, so maybe that lens wasn't so good (wider lenses are harder to make).  Maybe it wasn't at its sharpest F-stop.  A lot of the sharpest images across the frame grabs I looked at were close-ups, and I suspect back then a 50mm at it's sharpest aperture and focus distance was a lot better than a wider lens at whatever F-stop and focus distance was required for the scene.
Check out these grabs from 1955's The Seven Year Itch:


Monroe was at her height of popularity so there's no way she's getting the beat-up lenses from the rental company or a camera team that doesn't know what they're doing.  I can't think of any reason these two particular images would be softer than the technology at the time would have allowed.
By 1971 things seemed to have gotten a lot more consistent with Diamonds Are Forever:



Then if we fast forward to the last few years, we get films like 2024s Trap, which looks quite sharp and even approaching a digital look:


but still has films over-emphasis of high-contrast edges:

and it's a similar case with 2023's Poor Things which can look quite sharp:


but on wides it still has that film look:

and it's only when the lenses get crazy that the edges start looking more vintage again:

Anyway, lots of food for thought, but it's almost like the sweet spot of film has remained relatively similar in performance but has gotten drastically wider as you can now get images that are that sharp and clean in much less light and across a vastly wider range of lens focal lengths and apertures.
Another variable is that prior to digital projection, the final image had to go through many more layers of film than it has to now.  Back in the day the image pipeline was something like: negative → interpositive → internegative → release print, rather than just negative -> scan, and it's not like our projection lenses haven't gotten better now too! 
Let me know if you can think of any more variables that I didn't mention, but it's like we're looking at lenses get better and film be useful in more situations, rather than it get "better".

For those interested in small setups, in modestly priced gear, in non-clinical rendering of images, in very fast lenses, in vintage lenses, or older equipment, we exist in a space that has no quantitative reference.  There are no numbers to look up and understand things from.
It applies to the equipment:
Questions like "how sharp is that lens?" don't have an answer (that is intuitive anyway - MTF charts aren't intuitive and often aren't reliable or even available).  Even if it did, that answer would only be true at one aperture setting, and even then, is only true for the middle of the frame or the edge of the frame, but not both at the same time.
If we shoot at base ISO with a 4K camera then we'll likely get an image with roughly 4K resolution, but at higher ISOs the effective resolution will likely drop due to ISO noise, NR, compression, etc.
If we use filtration, like diffusion filters, then these lower the effective resolution of the image.  It's literally what they're designed to do.  How much do they do this though?  Not only is there no published answer to this, but the answer changes depending on focal length, sensor size, etc.
It applies to the look we're creating:
Any colourist working creatively will be trying to create an image with the right amount of resolution / sharpness / noise / etc, not just "the sharpest" or "the highest resolution".  How much is desired?  What are the references?
I've been struggling with many questions from my own equipment and projects, including:
My TTartisans 17mm F1.4 is less than half the weight of my Voigtlander 17.5mm F0.95, but the TT is quite soft at F1.4.  How soft is it though?  Is it too soft? If I want to shoot low-light with the GX85 (which has terrible higher-ISO NR) then what ISO setting is too soft, and then which lenses do I need to use in which environments to get enough light into the sensor? My Takumar 50mm F1.4 on my generic M42-MFT speed booster has pretty soft edges, but how soft? You'd think the solution to these would be to look at the footage and decide, but (for me at least) it's a double-edged sword because I also don't know what final images I want!  I have spent a good amount of time looking at Hollywood films and big budget TV shows (see the original The Aesthetic thread) but apart from just going "I like that" and "I don't like that" we have the problem once-again of there being no way to quantify things.  Saying "this show is softer than that show" really doesn't help.
My solution is to reference things back to film.
I originally did this with my Panasonic GF3, which shoots 1080p so soft you could cut it with a wet noodle, by comparing it to the presets in the Film Look Creator tool for 8mm and 16mm film I concluded that when the GF3 didn't show macro-blocking due to the (very low) bitrate, it was about the same as 8mm film.
This was actually a really useful reference for me, because the associations I have for 8mm and 16mm film are quite useful.  8mm film has an aesthetic that is very nostalgic and low-fi, but was never good enough for TV shows, let alone the cinema.
My new plan is to reference everything back to film, across quite a number of ways...
Texture, which is what I've talked about so far:
- I will be trying to "map" my lenses and cameras and codecs to a specific resolution of film (16mm, 35mm, somewhere in between)
- I will be trying to "map" my aesthetic preferences to film too, like wanting a certain project to have the resolution of 16mm for example, but further than this - the size and amount of grain can also be a useful reference.  These are useful references for me because a lot of the aesthetic references of cinema I have were actually shot on film and so by associating these things back to film it's a relevant reference, not just some arbitrary scale that isn't directly related.
Dynamic Range and Contrast:
- How does the DR from the GX85 look when put through an image pipeline in Resolve compare to the contrast of a 250D -> 2383 process?
- What about the iPhone vs a 16mm process from the 90s?  or a B&W process from the Italian Neorealism or French New Wave period?
Contrast and DR should be relatively easy to match to various film stocks by just shooting some over/under exposure tests and adjusting my standard Resolve colour pipeline to match what is in the spec sheets.
Speaking of spec sheets, not only do the spec sheets for motion picture film contain the Sensitometric Curves that show DR and contrast, but they also contain the MTF curves too as a reference for resolution.  When it comes to resolution you don't need to look at the charts though - I asked some film geeks I know to comment on the FLC presets and they said that the 8mm / 16mm / 35mm presets in the Grain panel have about the right amount of image softness and amount of grain (but that the character of the grain isn't accurate), so the FLC is a reasonable reference for the texture of film in a very broad sense.
What else?  
Image stability is another one.  8mm film cameras were larger than modern compact cameras so were more stable with the lenses they were normally fitted with, but 8mm had pretty terrible gate weave (alignment from one frame to the next) so having micro-jitters from hand-holding is compatible with the look.  Whereas 16mm would have had more mass and less gate weave but at least at first would have probably been shoulder mounted or on a tripod, so some types of shots / angles will be more compatible with the aesthetic than others.
Depth of field is another one.  Lots of people think the "Super 16mm look" just means deep DOF, but it's more nuanced than that, as the lenses typically used would have some separation in low-light when focused closer, but due to the lenses at the time the shots might have been softer wide-open, so that's another relationship to understand.
There are lots of other parameters that make an image that aren't covered here, but I am finding that getting some kind of reference for texture and contrast fills a very large gap in the landscape for me.  
The goal isn't to accurately emulate anything, its to develop a keener understanding of the spectrum these things exist in.
Where I'm hoping to get to is to be able to develop summaries like:
The GF3 is about 8mm at base-ISO, which during the day is equivalent to <some particular F-stop>, so I can put basically any lens sharper than 8mm onto it and the result will still look like 8mm.  I can hand-hold this tiny camera with an acceptable level of shake up to about Xmm and it'll still fit the 8mm vintage / amateur / nostalgic vibe.  The GF3 is tiny but once you add a lens that is larger than a pancake then I may as well use the GX85, so the only sensible lens is the 15mm F8 bodycap lens.  Any other combo doesn't make sense.
(This is an actual example I've worked out through testing).
  The GX85 at base-ISO is equivalent to <film size of some kind.. 16mm? 24mm? 35mm? 50mm?> which requires lenses of <F-stop> during the day and <F-stop> in well-lit night environments.  This amount of resolution is suitable for projects with a vibe of <gritty street? vintage? night cinema? high-end commercials? etc?> but not other vibes.
(This is still yet to be tested, but once I've worked out the camera then certain lens combinations will reveal themselves to make sense and others will obviously not work)
  iPhone?  Where does it sit in all this?  It has huge resolution and very strong codecs (4K Prores HQ or even Prores RAW) but poor DR and even worse ISO performance.
  GH7.  What are the aesthetics I want to create that I can't create with the above (because the above is too limiting).  What lenses and shooting styles and approaches are required for these aesthetics? The ultimate thinking is developing "constellations" where there is compatibility / alignment between: a camera, one or more lenses, certain shooting situations and techniques, an image pipeline, and a target aesthetic.  I've been working on finding these "constellations" by starting at the camera and working forwards, but also by starting with the end aesthetic and working backwards, and I've identified a number of partial matches, but I think that by relating everything back to motion picture film, I can make more progress fitting the pieces together.

Yes, indeed, but a few third-party brands offer that simple, inexpensive accessory, and it can easily be added to your setup.